In power converters, a PWM method has been widely used as a control method for obtaining a desired output voltage. The PWM method is a method of turning on switching devices for a time interval or duration proportional to a voltage command, and may be of a carrier comparison method or a space vector method as widely known in the art. The carrier comparison method may be used for controlling a general power converter with pulse with modulation, in which a desired pulse width is obtained by comparing a carrier signal with a voltage command value. The space vector method may be used for generating polyphase ac through a power converter such as an inverter, in which the ON duration of the switchg devices per unit time is determined based on a space vector representing voltage or magnetic flux.
The carrier comparison method, which is widely known in the art, is described, for example, at "5.2.3 PWM CONTROL" on pp 727-728 in "Electrical Engineering Handbook" published on Feb. 28, 1987 by I.E.E.J. The space vector method, which is also widely known, is described on pp 47-53 of "Theory and Design of AC Servo System" published on Aug. 19, 1991 by Sougou Electronics Publishing Co. Ltd. The details of these methods will not be describe herein.
With recent developments in the digital technology, digital hardware has been increasingly used in a PWM portion for the purpose of reducing the cost. In the digital hardware of the carrier comparison type, for example, instantaneous values of a voltage command signal and a carrier signal are treated as discrete values, and the carrier waveform and voltage command waveform are subjected to D-A (digital-to-analog) conversion, to provide stepped waveforms when displayed in an analog format. Here, the size (value) of one step, when expressed in terms of the output voltage, will be called "PWM resolution".
FIG. 6 shows the waveform of output voltage V.sub.out within one period of carrier signal when the resolution is d.sub.n, wherein the output voltage V.sub.out changes in steps for every value of d.sub.n. As the step size d.sub.n decreases, quantization errors occurring when obtaining discrete values can be accordingly reduced.
A carrier signal having a desired carrier frequency can be obtained, for example, by changing the number of counting a CPU clock with an updown counter. Where the carrier frequency is to be increased, the number of counting is reduced, and, consequently, the PWM resolution is reduced with a result of an increase in quantization errors. If the step size d.sub.n is sufficiently small and the output voltage is sufficiently large, resulting quantization errors are insignificant or trivial. Where a low output voltage is to be generated, particularly during low-speed rotation of a motor, voltage errors in the output voltage are increased, thus causing torque ripple or nonuniform rotation.
FIG. 7 shows a known technique of reducing errors in the output voltage.
In order to reduce errors in the output voltage, output voltage detecting means 12 detects the output voltage, and the detected voltage and a command value for the output voltage are fed to a voltage regulator (AVR) 11 so as to correct a command value for generating PWM pulses, thereby to provide a desired output voltage. Here, the PWM method is of a carrier comparison type in which a carrier generating means 5 and a comparator 6 are used.
The use of feedback control as shown in FIG. 7 in order to reduce quantization errors results in an increase in the cost due to the provision of a sensor portion including the voltage detecting means. In particular, the voltage detecting means needs to be insulated from a main circuit, and thus requires an expensive insulating amplifier, with a result of a significant increase in the cost. Also, an expensive CPU is needed in the method in which the resolution d.sub.n is reduced by increasing the frequency of the CPU clock.